vibencode
Case study

Sleek Email

An email client that groups a mailbox by sender and clears the routine half before anyone reads a line.

Client Sleek Sector Productivity software
The shape of it The sentence it arrived as
What we heard

“People lose an hour a day in here. Can it sort itself out?”

What it became

Grouping by sender across the mailbox, rules that raise and file unasked, and a model that learns what each person opens.

What we found The problem under the stated one

Every mail client treats the message as the unit. Almost nobody thinks that way.

People think in senders: this vendor, that client, the build server. A sender’s mail gets dealt with in one sitting or not at all, and a list ordered by arrival time cuts across the only grouping the reader is using.

Filters and labels were never missing from these products. They go unused because configuring them is a job, and the people drowning have no afternoon to spend on it. So the thing to automate was not the sorting. It was the decision to sort.

An inbox rule that has to be set up is a rule that never gets set up.

Which is why nothing in the sorting asks the reader a question first.
What we built In the order it happened
  1. 01Group by sender and domain, so everything from one place arrives as a single thing to deal with.
  2. 02Rules that act rather than label: raise what matters, file what does not, hold what is waiting on a reply.
  3. 03A model over what each person opens, answers and ignores, so the rules sharpen without anyone writing one.
  4. 04The pass runs ahead of the first read, so the mailbox someone opens is already the short version.
Ships with
  • The sender graphCorrespondents scored on how the account has actually treated them.
  • The rule engineActions rather than labels, applied before anyone opens the app.
  • The per-user modelTrained on opens, replies and deletions, so it moves with the person.
What it does now In production

Mail arrives grouped, the routine half already filed, and what is left is the people waiting on a reply.

The rules sharpen on their own: the model watches what gets opened, answered and deleted, and the grouping follows.

Taken over
  • The daily triage pass
  • Writing and maintaining filter rules
  • Chasing what is waiting on a reply

This one sits across GenAI implementation and Agents and orchestration. Most engagements touch two.

Other engagements

Is yours the same shape as this one?

Book a 30-minute call

Describe it in your own words. We will tell you where it lands.